Type: article
Abstract: Knowledge workers increasingly use generative AI in analytical tasks, but high textual overlap with AI outputs does not necessarily indicate uncritical dependence. We propose that behavioral text adoption and self-reported reliance on AI can be combined as a diagnostic lens for distinguishing different forms of AI-assisted work. In a study of 102 business students completing a competitor analysis with access to a conversational AI assistant, more proficient participants were less likely to use the assistant in a single-shot manner and more likely to engage iteratively. Higher text adoption was associated with higher-quality outputs, but this association weakened as self-reported reliance increased. This pattern was statistically significant under both human and LLM-based evaluation. These findings suggest that the same degree of AI text adoption may reflect different human-AI processes and suggest caution in treating text overlap alone as a marker of overreliance.
Institution: NOVA School of Business and Economics
Language: English
Topic: AI and HR Technologies
Domain: Social Sciences
Open Access status: green
OpenAlex ID: W7222988653